Subway passenger flow volume influence factor analysis method

Abstract

The invention belongs to the technical field of urban traffic data analysis, and particularly relates to a subway passenger flow volume influence factor analysis method which comprises the following steps: extracting effective information of original data; three latitudes of density, land utilization diversity and street design are selected to represent subway station construction environment characteristics; selecting a GTWR model to explore the influence of the independent variable on the subway passenger flow volume; oLS, GWR and GTWR models are compared through AcrGIS software, the GTWR model is verified to have the optimal effect, and the GTWR model is most suitable for capturing the space-time relation between the explanatory variable and the daily hour passenger flow volume; carrying out descriptive statistical analysis by utilizing a GTWR model; variables with obvious time and space heterogeneity are selected from the result of the GTWR model for discussion. The defects in the prior art are overcome, and an urban planner is helped to formulate an effective and reliable strategy for rail transit transfer service by considering multiple last kilometer access modes.

Description

Technical Field

[0001] The invention belongs to the technical field of urban traffic data analysis, and in particular relates to a method for analyzing factors affecting subway passenger flow. Background Art

[0002] The rapid increase in the number of private cars has caused a series of social problems, such as traffic congestion and environmental pollution. As an effective way to solve these problems, public transportation priority policies have attracted widespread attention from scholars. With its large carrying capacity, energy saving, environmental protection, speed and efficiency, subways have become a key component of urban public transportation systems. However, compared with private cars, subways still cannot achieve a door-to-door service model, and often need to connect with other modes of transportation to solve the last mile problem, such as park-and-ride, bicycle transfers, and bus transfers. The last mile facilities have been considered to be a key factor in improving subway accessibility and have a significant impact on the total travel time of subway passengers. Therefore, understanding how the last mile facilities affect subway passenger flow is crucial for rail transit planning and urban development policy formulation.

[0003] At present, existing studies have confirmed that the first / last mile facilities affect the travel choices and overall riding experience of subway passengers. Specifically, good connecting facilities such as bus stations, parking lots, bicycle facilities and shared bicycle services can significantly increase the utilization rate of subways and improve the commuting convenience of passengers. A large number of studies have explored the impact of last mile facilities on subway passenger flow, and direct ridership models (DRMs) based on regression methods have been widely used. Ordinary least squares (OLS) regression was widely used in early studies, but it cannot explore the spatial heterogeneity of variables.

[0004] Therefore, the geographically weighted regression (GWR) model that can overcome the spatial heterogeneity of variables has been proposed. Existing studies have used the GWR model to consider subway passenger flow during peak hours to explore the temporal impact of explanatory variables on subway passenger flow. However, they cannot reflect the temporal fluctuations of the impact of explanatory variables on subway passenger flow within a day. Summary of the invention

[0005] The purpose of the present invention is to provide a method for analyzing factors affecting subway passenger flow, which overcomes the shortcomings of the prior art and comprehensively analyzes the spatiotemporal heterogeneous relationship between subway passenger flow and influencing factors by using the last-mile facilities, built environment characteristics and station characteristic data through a temporal geographic weighted regression model.

[0006] To solve the above problems, the technical solutions adopted by the present invention are as follows:

[0007] A method for analyzing factors affecting subway passenger flow, comprising the following steps:

[0008] S1. Extract subway passenger flow information, including the daily average passenger flow of subway stations, the spatial and temporal distribution of passenger flow, and the peak passenger flow time;

[0009] S2, taking the area within 500 meters of the subway station as the buffer zone, extracting the independent variables within the buffer zone;

[0010] S3. Construct a GTWR model that reflects the impact of independent variables on subway passenger flow. The expression of the GTWR model is:

[0011] y θ =α0(μ θ ,v θ ,t θ )+∑α k (μ θ ,v θ ,t θ )x θk +ε θ

[0012] In the formula, (μ θ ,v θ ,t θ ) represents the geographical coordinates and observation time of the centroids of θ subway stations, (μ θ ,v θ ) is the spatial coordinate of the θth subway station, t θ is the observation time, α0(μ θ ,v θ ,t θ ) is the intercept, ε θ is the Gaussian error; α k (μ θ ,v θ ,t θ ) is represented by the θth station at t θ The estimated values ​​of regression parameters of different variables at the time;

[0013] S4. Using R 2 , AICc and RSS were used to evaluate the performance of the GTWR model;

[0014] S5. Based on the output results of the GTWR model, the influence of the explanatory variables on the subway passenger flow is described according to the quartiles, medians and maximum values, and the descriptive statistics of the regression coefficients are obtained to analyze the influence of different independent variables on the subway passenger flow;

[0015] S6. Analyze the factors affecting subway passenger flow from the perspectives of time and space.

[0016] Furthermore, the independent variables described in S2 include road density, office and entertainment venue density, land mix, subway station type, number of bus stations and bus routes, number of shared bicycle trips, number of bicycle lanes, greening rate, and number of parking lots.

[0017] Furthermore, in S3 k (μ θ ,v θ ,t θ ) in (μ θ ,v θ ,t θ The expression of the local estimated weighted least squares estimate at ) is:

[0018]

[0019] In the formula, δ(μ θ ,v θ ,t θ ) is the spatiotemporal weighting matrix, which is an n×n matrix and its expression is:

[0020] Q i =δ(μ θ ,v θ ,t θ )=diag(δ θ1 ,δ θ2 ,...δ θn )

[0021] Where Y is the daily hourly passenger flow matrix, and the matrix δ(μ θ ,v θ ,t θ ) represents the spatiotemporal weight matrix at the grid θ, and the weight is calculated based on the Gaussian distance decay function, and its expression is:

[0022]

[0023] In the formula, δ θj represents the weight between station θ and station j, b DT represents the space-time bandwidth;

[0024] represents the space-time distance, and its expression is:

[0025]

[0026] Furthermore, considering that place and time usually have different scale effects, the temporal distance is converted into an additional spatial scale distance using the spatiotemporal scale factor. The expression of the GTWR model is:

[0027] y θ =α0(μ θ ,v θ ,t θ )+[X T Q θ (μ θ ,v θ ,t θ )X] -1 X T Q θ (μ θ ,v θ ,t θ )Y+ε θ .

[0028] Furthermore, the factors affecting subway passenger flow are analyzed from the perspectives of time and space as described in S6, including:

[0029] (1) Coefficient time distribution: Variables with more obvious time heterogeneity in the results of the GTWR model are selected for discussion; the average value of the regression coefficient of the explanatory variable is calculated at hourly intervals, the impact of the explanatory variable on the subway station passenger flow is analyzed from the time dimension, and the impact of each variable on the subway passenger flow is obtained from the time distribution characteristic diagram of the regression coefficient;

[0030] (2) Spatial distribution of coefficients: We select significant variables such as the number of bus stops, the number of parking lots, the density of shared bicycle companies, and the number of transfer lines to analyze the spatial differences in their impacts; we draw a distribution map of the average hourly coefficients during peak hours to reflect the impact of last-mile facilities, built environment elements, and station attributes on subway passenger flow.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] 1. This paper uses daily passenger flow data and the GTWR model to explore whether and how the last-mile facilities have an impact on subway passenger flow, and whether there is spatiotemporal heterogeneity in the interactive relationship between the last-mile facilities and subway passenger flow.

[0033] 2. Compared with the more widely used GWR model, the present invention adopts the GTWR model, which not only takes into account the spatial heterogeneity, but also takes into account the non-stationarity of the passenger flow of rail transit stations in the temporal dimension, which is affected by peak and non-peak hours.

[0034] 3. The analysis method provided by the present invention uses a wide range of data and has high accuracy. Density, land use diversity, and street design are selected as three dimensions to characterize the built environment characteristics of subway stations. Three common last-mile transportation modes, including buses, bicycles, and cars, are considered to evaluate the service level of their last-mile transportation facilities, and the types of independent variables are more comprehensive.

[0035] 4. The present invention helps urban planners to formulate effective and reliable strategies for rail transit transfer services by considering various last-mile access modes, and provides strong data support for transfer management. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of the process of the present invention.

[0037] Figure 2 This is the spatial distribution map of passenger flow in the research area of ​​the present invention.

[0038] Figure 3 This is the time distribution histogram of passenger flow in the study area of ​​the present invention.

[0039] Figure 4 It is the time distribution characteristic diagram of the average regression coefficient of the model of the present invention.

[0040] Figure 5 It is the spatial distribution diagram of the average hourly coefficient during the peak period of the model of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] The following is combined with Figure 1 The technical solution of the present invention is specifically described. This patent relates to a method for analyzing the temporal and spatial impact of built environment factors and last-mile facilities on subway passenger flow, including the following steps:

[0043] Step 1: Select the research area and object

[0044] The research area selected in the present invention is Shanghai, and the passenger flow data of 297 stations on 16 lines of Shanghai Metro are selected, and the data comes from Shanghai Metro Company.

[0045] Step 2: Current passenger flow analysis

[0046] The present invention analyzes the passenger flow data of the selected 297 stations from four aspects: the daily average passenger flow of the stations, the peak time of passenger flow, the spatial distribution of passenger flow and the temporal distribution of passenger flow.

[0047] The spatial distribution of passenger flow is shown in the attached figure. Figure 2 As shown; draw a histogram to show the time distribution of passenger flow, as shown in the attached Figure 3 shown.

[0048] Step 3: Extracting independent variables

[0049] All independent variables in this invention use a 500m buffer zone as the spatial range of measurement, and select density, land use diversity, and street design as three dimensions to characterize the built environment characteristics of subway stations. In this invention, company density and entertainment and leisure density are used to characterize the density dimension; road density and other indicators are used to characterize the design dimension; and land use mix degree is used to characterize diversity. POI data is extracted from the API (Application Programming Interface) interface open to Amap. The specific calculation method is as follows:

[0050] (1) Road density

[0051] Road density was selected as the indicator of street design. The data was obtained from the OpenStreetMap open source map, and the road network density was calculated using the ArcMap platform.

[0052] (2) Company density and entertainment density

[0053] The company density and recreation density were calculated as the ratio of company POIs and recreation POIs to the total number of 13 types of POIs within the 500-m buffer zone, respectively.

[0054] (3) Land Mixture

[0055] Land use mixture is an entropy index that measures the degree of mixture of different land use types. The formula is as follows:

[0056]

[0057] Where, Ei is the land use entropy index of subway station; A ip is the proportion of the pth POI type in subway station i; i is the number of POI types contained in subway station i.

[0058] (4) Station type

[0059] The number of transfer lines for a subway station (e.g. People's Square Station is a transfer station for Metro Lines 1, 2, and 8, with the number of transfer lines = 3) and whether it is a terminal station are used as indicators of subway station type. The data are obtained through the Shanghai Transportation Network.

[0060] (5) Number of bus stops and routes

[0061] The number of bus stops and bus routes within the buffer zone of subway stations are used to evaluate the intermodal connection service level between subways and urban bus systems. The data are extracted through the API interface.

[0062] (6) Number of shared bicycle trips

[0063] The shared bicycle data from 6:00 to 23:00 on weekdays was selected, and a buffer zone was established within 50m of the subway entrance and exit. The number of shared bicycle trips that started or ended was used as an indicator of shared bicycle service.

[0064] (7) Availability of bicycle lanes

[0065] Whether there is a bicycle lane within the 500m buffer zone is selected as an indicator for evaluating the level of bicycle transfer service.

[0066] (8) Green space ratio

[0067] Green space data is drawn based on Google Earth images and urban boundary data, and the green space ratio is the ratio of green space area to buffer zone area.

[0068] (9) Number of parking lots

[0069] The number of registered parking lots within a 500m buffer zone was selected to evaluate the park-and-ride service level.

[0070] Step 4: Model selection

[0071] The GTWR model can reflect the spatial and temporal heterogeneity of each variable on subway passenger flow when explaining variables, and has higher accuracy than the OLS model and GWR model. Based on this, the present invention uses the GTWR model to explore the impact of independent variables on subway passenger flow.

[0072] The GTWR model is described as:

[0073] y θ =α0(μ θ ,v θ ,t θ )+∑α k (μ θ ,v θ ,t θ )x θk +ε θ (2)

[0074] In the formula, (μ θ ,v θ ,t θ ) represents the geographical coordinates and observation time of the centroids of θ subway stations, (μ θ ,v θ ) is the spatial coordinate of the θth subway station, t θ is the observation time, α0(μ θ ,v θ ,t θ ) is the intercept, ε θ is the Gaussian error; α k (μ θ ,v θ ,t θ ) is represented by the θth station at t θ The estimated values ​​of regression parameters of different variables at the time.

[0075] α k (μ θ ,v θ ,t θ ) in (μ θ ,v θ ,t θ The expression of the local estimated weighted least squares estimate at ) is:

[0076]

[0077] In the formula, δ(μ θ ,v θ ,t θ ) is the spatiotemporal weighting matrix, which is an n×n matrix and its expression is:

[0078] Q i =δ(μ θ ,v θ ,t θ )=diag(δ θ1 ,δ θ2 ,...δ θn )

[0079] Where Y is the daily hourly passenger flow matrix, and the matrix δ(μ θ ,v θ ,t θ ) represents the spatiotemporal weight matrix at the grid θ, and the weight is calculated based on the Gaussian distance decay function, and its expression is:

[0080]

[0081] In the formula, δ θj represents the weight between station θ and station j, b DTrepresents the space-time bandwidth;

[0082] represents the space-time distance, and its expression is:

[0083]

[0084] Considering that place and time usually have different scale effects, we use the spatiotemporal scale factor to convert the temporal distance into an additional spatial scale distance.

[0085] Combining the above model and derivation results, we substitute the formula into formula (2) to obtain a comprehensive summary formula, which can be expressed as:

[0086] y θ =α0(μ θ ,v θ ,t θ )+[X T Q θ (μ θ ,v θ ,t θ )X] -1 X T Q θ (μ θ ,v θ ,t θ )Y+ε θ (6)

[0087] y θ is the influence coefficient of each independent variable on different subway stations at different times, Q θ is the spatiotemporal matrix of regression parameter estimation, ε θ is a constant term that contains random factors not explained in the model.

[0088] Step 5: Model Results Evaluation

[0089] The present invention utilizes R in this step 2 , AICc and RSS are used to evaluate the performance of the GTWR model. 2 It represents the goodness of fit of the model. The larger the value, the better the model explains the dependent variable. The smaller the AICc value and RSS value, the better the explanatory power of the model. Based on these three indicators, the GTWR model is compared with the OLS model and the GWR model, and finally it is verified that the GTWR model is suitable for this patent.

[0090] Step 6: Descriptive statistical analysis of regression coefficients

[0091] The present invention selects quartiles, medians and extreme values ​​to describe the influence of explanatory variables on subway passenger flow, obtains a descriptive statistical table of regression coefficients, and finds that the same explanatory variable has different directions and influence degrees on passenger flow at different rail transit stations at different times.

[0092] Step 7: Analysis of spatiotemporal heterogeneity of passenger flow based on model results

[0093] From the results of the GTWR model, the average value of the regression coefficient of the explanatory variable is calculated at hourly intervals, the impact of the explanatory variable on the passenger flow of the subway station is analyzed from the time dimension, and the time distribution characteristic diagram of the regression coefficient is drawn, as shown in the attached figure. Figure 4 shown.

[0094] Due to space limitations, significant variables such as the number of bus stops, the number of parking lots, the density of shared bicycles and companies, and the number of transfer lines were selected to analyze the spatial differences in impact. ArcGIS was used to draw the spatial distribution map of the average hourly coefficient during the morning and evening peak hours to reflect the impact of different variables on subway passenger flow, as shown in the attached figure. Figure 5 shown.

[0095] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for analyzing factors affecting subway passenger flow, characterized by: The following steps are involved: S1. Extract subway passenger flow information, including the daily average passenger flow of subway stations, the spatial and temporal distribution of passenger flow, and the peak passenger flow time; S2, taking the area within 500 meters of the subway station as the buffer zone, extracting the independent variables within the buffer zone; S3. Construct a GTWR model that reflects the impact of independent variables on subway passenger flow. The expression of the GTWR model is: y θ =α0(μ θ ,v θ ,t θ )+Sa k (m θ ,v θ ,t θ )x θk +e θ In the formula, (μ θ ,v θ ,t θ ) represents the geographical coordinates and observation time of the centroids of θ subway stations, (μ θ ,v θ ) is the spatial coordinate of the θth subway station, t θ is the observation time, α0(μ θ ,v θ ,t θ ) is the intercept, ε θ is the Gaussian error; α k (μ θ ,v θ ,t θ ) is represented by the θth station at t θ The estimated values ​​of regression parameters of different variables at the time; S4. Using R 2 , AICc and RSS indicators were used to evaluate the performance of the GTWR model; S5. Based on the output results of the GTWR model, the influence of the explanatory variables on the subway passenger flow is described according to the quartiles, medians and maximum values, and the descriptive statistics of the regression coefficients are obtained to analyze the influence of different independent variables on the subway passenger flow; S6. Analyze the factors affecting subway passenger flow from the perspectives of time and space.

2. The method for analyzing factors affecting subway passenger flow according to claim 1, characterized in that: The independent variables described in S2 include road density, office and entertainment venue density, land mix, subway station type, number of bus stations and bus routes, number of shared bicycle trips, number of bicycle lanes, greening rate, and number of parking lots.

3. The method for analyzing factors affecting subway passenger flow according to claim 1, characterized in that: S3 α k (μ θ ,v θ ,t θ ) in (μ θ ,v θ ,t θ The expression of the local estimated weighted least squares estimate at ) is: In the formula, δ(μ θ ,v θ ,t θ ) is the spatiotemporal weighting matrix, which is an n×n matrix and its expression is: Q i =δ(μ θ ,v θ ,t θ )=diag(δ θ1 ,d θ2 ,...d θn ) Where Y is the daily hourly passenger flow matrix, and the matrix δ(μ θ ,v θ ,t θ ) represents the spatiotemporal weight matrix at the grid θ, and the weight is calculated based on the Gaussian distance decay function, and its expression is: In the formula, δ θj represents the weight between station θ and station j, b DT represents the space-time bandwidth; represents the space-time distance, and its expression is:

4. The method for analyzing factors affecting subway passenger flow according to claim 3, characterized in that: Considering that place and time usually have different scale effects, the temporal distance is converted into an additional spatial scale distance using the spatiotemporal scale factor. The expression of the GTWR model is: y θ =α0(μ θ ,v θ ,t θ )+[X T Q θ (μ θ ,v θ ,t θ )X] -1 X T Q θ (μ θ ,v θ ,t θ )Y+ε θ 。 5. The method for analyzing factors affecting subway passenger flow according to claim 1, characterized in that: S6 analyzes the factors affecting subway passenger flow from the perspectives of time and space, including: (1) Coefficient time distribution: Variables with more obvious time heterogeneity in the results of the GTWR model are selected for discussion; the average value of the regression coefficient of the explanatory variable is calculated at hourly intervals, the impact of the explanatory variable on the subway station passenger flow is analyzed from the time dimension, and the impact of each variable on the subway passenger flow is obtained from the time distribution characteristic diagram of the regression coefficient; (2) Spatial distribution of coefficients: We select significant variables such as the number of bus stops, the number of parking lots, the density of shared bicycle companies, and the number of transfer lines to analyze the spatial differences in their impacts; we draw a distribution map of the average hourly coefficients during peak hours to reflect the impact of last-mile facilities, built environment elements, and station attributes on subway passenger flow.

Images